It is known in the prior art to create inference engines based on data and inferences about each datum (also referred to herein as “training data”). When an inference engine processes a new datum, the engine determines an inference based on the previously processed training data.